Germany’s armed forces have started testing loitering munitions and AI-assisted targeting tools in Lithuania, close to the Belarusian border, in a live military exercise that says as much about Europe’s procurement mood as its battlefield planning.
Euronews, citing Reuters, reported that the trials are part of the multinational Freedom Shield exercise, involving about 3,800 personnel from eight NATO countries at the Rūdninkai training area. The systems under test include drones from German defence technology groups Helsing and Stark Defence, along with Helsing’s Clarity targeting platform.
The hardware is not subtle. So-called kamikaze drones are built to loiter, watch and then destroy themselves on impact once an attack is authorised. In Lithuania, German troops used inert warheads against simulated armoured targets, which limits what the exercise can prove about destructive effect but still allows the military to test guidance, coordination and operator workflow. That is the part modern armies keep buying: not just explosives, but the speed between seeing something and deciding what to do about it.
That matters because the denominator in Europe’s defence debate has changed. A tank, artillery system or supply convoy is expensive, scarce and slow to replace. A drone designed to threaten it is comparatively cheap. Ukraine has made that arithmetic impossible to ignore, even if it has also shown the other side of the ledger: attack drones can be jammed, disrupted and swatted down by counter-drone systems once the enemy adapts. Cheap does not mean unstoppable. It means attrition gets democratised.
Germany’s choice of Lithuania is strategic rather than theatrical. The country sits near both Belarus and Russia’s Kaliningrad region, and Berlin has already committed to stationing Panzerbrigade 45 there, with full operational capability expected by 2027 and roughly 5,000 military personnel and supporting civilian staff. Testing drones and AI tools where that brigade is meant to operate is less a tech demo than a rehearsal for a more permanent posture.
The AI layer is where the ambition, and the risk, sits. Helsing’s software is designed to process inputs from drones and other sensors to help identify targets and speed decisions. Military organisations love any product that promises to compress time. The difficulty is that faster recommendation is not the same thing as better judgment, especially when the system is sorting battlefield signals rather than invoices. Armies, like investors, tend to discover model risk after deployment.
Still, Germany is not experimenting in a vacuum. It is trying to work out whether software, sensors and expendable aircraft can give expensive brigades a better chance of surviving first contact.
Isla Camilleri
Alex de Valletta
Dua Mifsud
Ryan C